Is granite-4.0-h-350m or embeddinggemma-300m smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

granite-4.0-h-350m — its KV cache at 32K context is 1.2x smaller, because of how its attention layers cache.

Model comparison

granite-4.0-h-350m vs embeddinggemma-300m

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: granite-4.0-h-350m's KV cache at 32K is 1.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	granite-4.0-h-350m	embeddinggemma-300m
Parameters	340M	303M
Architecture	granitehybrid	gemma-embedding
Layers	32	24
Native context	32,768	2,048
Mixture of experts	no	no
Quantizations published	29	10
Smallest quantization	0.15 GiB	0.26 GiB
Q4_K_M	0.21 GiB	—
Licence	apache-2.0	—

KV cache by context

the term that decides long-context viability

Context	granite-4.0-h-350m	embeddinggemma-300m	Ratio
4,096	0.02 GiB	0.04 GiB	2.25×
8,192	0.03 GiB	0.05 GiB	1.62×
16,384	0.06 GiB	0.08 GiB	1.31×
32,768	0.13 GiB	0.14 GiB	1.16×
65,536	0.25 GiB	0.27 GiB	1.08×
131,072	0.50 GiB	0.52 GiB	1.04×